Comments on Sylvia Berryman, "Ancient Automata and Mechanical Explanation"
Tim O’Keefe
25th Annual Workshop in Ancient Philosophy, University of Texas at Austin, March 2002.
[A revised version of Sylvia Berryman’s paper was subsequently published in Phronesis, Volume 38, #4, 2003]
I’m happy to have the opportunity to comment on Sylvia’s paper. This paper is only one part of a larger project that Sylvia is working on on ‘mechanistic’ thought in the ancient world. I think it’s exciting, innovative, and important work. I look forward to seeing all of this work in print soon.

I'd first like to summarize briefly what I think are Sylvia's main claims and her reasons for making them, then I'll move on to a few questions and criticisms.

Sylvia makes two claims: the first about the content of ancient natural philosophy, and the second, about the technology of those times, that is supposed to explain (at least partly) why the first claim is true.

  1. Philosophical claim. No philosopher prior to the late Hellenistic period had--properly speaking--a "mechanical conception" of nature.
  2. Technological claim. (1) is the case because none of the devices described in early scientific literature could have served as the inspiration for "mechanical conceptions" of nature.
On Sylvia’s view, To have a "mechanical conception" of nature is to draw analogies/comparisons between various natural processes–such as animal self-motion–and devices using techniques of the mechanics of the time, and to understand the natural processes as operating like these devices.

Sylvia supports claim (2) by examining purported ancient accounts of automata. Since one of the most important features of animals is that they're "self-movers," a "mechanical conception" of nature would include making comparisons between organisms and working artifacts that are automata (or imaginary device that use only techniques essentially available at the time), and understanding the workings of organisms by reference to these devices.

Sylvia utilizes the following characterization set out by Aristotle of a self-mover:

"A self-mover need not initiate motion in the absence of any other changes; however, it is not directly operated by an agent, and the construction of the device is such as to transform the initial input into a motion of a different kind."

This seems to set up two necessary conditions on something counting as an automaton:

  1. it is not directly operated by an agent,
  2. because of its construction--that is, because of the way its parts are put together-- it is able to transform the initial input into motion of a different kind.
A question: what is meant by the ability to "transform the initial input into motion of a different kind"? Sylvia gives the example that a projectile would not count, but what would count is less clear. Does this mean transforming qualitative change into local motion…? How about releasing a catch on a wind-up catapult that causes the arm to move forward…?

Then, with these two conditions in hand, Sylvia debunks various purported examples of automata that pre-date Aristotle. She primarily relies on the italicized part of condition (ii): the motion of these devices that replicate animal self-motion is not understood to occur because of the internal construction of the bits and pieces making it up. Instead, this motion is explained by reference to things like:

Sylvia then goes on to map out a three-stage historical process, describing the availability of automata in ancient technology, and the philosophical uses of mechanistic analogies:

Three-stage historical process:

(A) Pre-Aristotle, there are no clear reports of ancient automata that satisfy the dual criteria for automata, and no thinkers conceive of natural processes in 'mechanistic' terms.

(B) The self-moving puppets Aristotle discusses do fulfill the dual criteria, but Aristotle himself does not conceive of organisms as analogous to these automata. He is simply using the analogy to illustrate a single point about animal motion. None of Aristotle's contemporaries or immediate successors picks up on this type of analogy in order to conceive of animal self-motion in "mechanistic"/automaton terms.

(C) It is only much later in the Hellenistic period that the wide availability of mechanical and pneumatic devices is attested to, and that some thinkers started to conceive of animal motion in 'mechanistic'/automaton terms. We see attempts to explain particular bodily functions in "mechanistic" terms in the schools of Erasistratus (3rd century BCE) and Asclepiades (c.129 - 40 BCE). The first evidence of a thoroughgoing 'mechanistic' conception of organisms is in Galen's (c. 129-210 CE) attack on the 'third-way' philosophers; later still we find the position that Plotinus attacks, that construction techniques can explain the natural world.

Sylvia also makes a terminological recommendation about the use of the term 'mechanistic.'

Terminological recommendation: Because ancient thinkers did not conceive of animal self-motion in "mechanistic"/automaton terms, and did not use comparisons to techniques from mechanics of the day (or conceivable extensions of it) in order to explain natural processes, we should not continue to call these ancient thinkers--especially the atomists--'mechanistic' thinkers (in contrast to 'teleological' thinkers like Aristotle and the Stoics). This usage is anachronistic and misleading. Instead, we should reserve the label 'mechanistic' for those systems of thought that do use such analogies.

Turning now to my evaluation/comments.

Two points that I think Sylvia is absolutely right about, and they're important, valuable points:

However, although I'm sympathetic to Sylvia's main claim that technological advances had a major impact on how thinkers could conceive of nature, in my role as commentator, let me raise a few questions about her account.

First, about period A, which occurs before Aristotle. Sylvia wants to claim that none of the devices described in early literature could have inspired a 'mechanistic conception.' However, what she has shown is that many of the 'self-moving' devices described in early reports couldn't have served as mechanical models of animal self-motion. The latter claim is narrower than the former. Couldn't there be mechanistic analogies without there being full-fledged automata available? I wonder whether some of the devices that were available before Aristotle, such as catapults and wine presses, could have been used for such machine analogies. A catapult, for instance, might be able to satisfy condition (ii) of the dual conditions for an automaton, depending on what is meant by motion "different in kind." In which case even if Archytas' 'Dove' were a catapult, it could go partway in modeling self-motion. That is, I think that some of devices actually available at the time did fall somewhere between full-fledged automata and imaginary 'devices' powered via magic, or quite simple devices like projectiles. And perhaps these devices could have served as a basis for asking the question of whether nature works 'like a machine.' If someone objects that such devices couldn't be used to explain all of the features that nature has--such as animal self-motion--I'd reply that this is still true even of the more sophisticated later devices that Sylvia describes, since (for example) there is nothing about such devices that would explain why animals reproduce within their kinds. It's a further question whether these devices did spur the question whether nature is like a machine, but all I'm asking is whether Sylvia is right that there weren't devices available at the time that could have spurred that sort of question.

My second point concerns stage (C) of the historical process Sylvia describes. I'm uncertain whether the reports from Galen and Plotinus are full enough to be able to say much about the emergence of a new, ‘mechanistic’ way of understanding animal motion. Both the Galen and Plotinus passages mention ‘mechanistic’ understandings of natural phenomena only in passing, The people Galen describes do compare the workings of organisms to those of theatrical devices. But can we assert confidently that they assert (as Sylvia puts it) that there is "nothing different in kind" between the functioning of organisms and the operations of mechanical devices? And even if these thinkers do have a full-blown 'mechanistic' philosophy, this way of conceiving of nature seems not to be widespread in the ancient world, if we only have it reported in Galen's "On The Formation of the Fetus." In the case of Plotinus, it's not clear to whom he's referring when he says that the workings of nature are often compared to those of doll-makers. He may be attacking a well-developed mechanistic position of a rival school. But he may merely be trying to spell out more precisely in what sense nature is (and is not) a craftsman. He may even have Aristotle's discussion in De Motu Animalium 7 in mind.

Similarly, Sylvia cites the attempts by the schools of Erasistratus and Asclepiades to account for the movement of fluids throughout the body by analogy to pneumatic technology as evidence that new technology led thinkers to conceive of organisms in 'mechanistic' ways. Yet earlier, Sylvia discusses the attempt in a Hippocratic text to explain the sorting of materials in a body by comparing this to blowing through a bladder . She says that this does not count as a 'mechanistic conception' of nature, since this text uses the analogy to explain only the functioning of a particular organ . "A ‘mechanistic conception,’ by contrast, takes technological means to be sufficient for replicating the functions of organisms; living things contain nothing different in kind." But by those standards, why would the explanations of fluid movement by the schools of Erasistratus and Asclepiades through the body count as 'mechanistic conceptions'? It seems, at least initially, that either both the Hippocratic and later explanations should count as mechanistic conceptions, or neither should.

One final point. Sylvia writes: "if the root sense of ‘mechanistic’ refers to a method, not a particular set of contents—a method of applying the techniques of ‘mechanics’—then the nonanachronistic sense of a ‘mechanistic conception’ would be one that uses ideas from the mechanics of the day to understand nature." Her point is well-taken, but I'm not certain I agree with it. After all, the meaning of terms aren't set by their etymological roots, but instead by their usage. And I think that the term 'mechanistic,' at least within the philosophical community today, has acquired a meaning through its usage that's quite far from the root sense of machine analogies. Although it may not have a single well-defined meaning, 'mechanistic' thought seems to involve some combination of a rejection of teleological explanation and of Aristotelian substantial forms and other so-called 'occult' properties, and a commitment to explaining the properties and behaviors of bodies from the "bottom up" by referring to the bits and pieces of matter that make them up. (NB: this isn't meant to be a rigorous definition; for now I'm just gesturing at how I think the term is actually used.) It's useful to have a term to refer to this sort of position, and I'd be reluctant to give it up unless Sylvia has a suggestion about a better replacement term. In order to argue that we should give up the term 'mechanistic' as currently used, I don't think it's enough to note that it might be anachronistic and strays from the root sense. I think Sylvia would need to argue that that the use of the term is positively misleading and confusing, or that the various ways that it's used today are so varied, muddled, and perhaps mutually inconsistent--as is the case with the term "post-modern"--that it's better just to scrap the term, at least as currently used.

Back to the list of Tim's publications and presentations